CN110955834A - Knowledge graph driven personalized accurate recommendation method - Google Patents
Knowledge graph driven personalized accurate recommendation method Download PDFInfo
- Publication number
- CN110955834A CN110955834A CN201911183454.6A CN201911183454A CN110955834A CN 110955834 A CN110955834 A CN 110955834A CN 201911183454 A CN201911183454 A CN 201911183454A CN 110955834 A CN110955834 A CN 110955834A
- Authority
- CN
- China
- Prior art keywords
- knowledge
- user
- node
- graph
- article
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/042—Knowledge-based neural networks; Logical representations of neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/02—Knowledge representation; Symbolic representation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Databases & Information Systems (AREA)
- Business, Economics & Management (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Animal Behavior & Ethology (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention provides a knowledge graph driven personalized accurate recommendation method, which comprises the steps of obtaining related knowledge of an article from a knowledge base according to historical behaviors of a user, constructing a knowledge graph, initializing vector representation of each node and connection, and determining a perception domain of the node; generating a training sample according to the historical behaviors of the user, and initializing vector representations of all users and articles; acquiring a receptive field of an entity corresponding to an article in a knowledge graph in a training sample, and inputting the receptive field and the sample as a graph neural network model to obtain a predicted value of the possibility of interaction between a user and the article; optimizing the model parameters by minimizing a loss function; and after the model optimization process is finished, sequencing the possibility predicted values of interaction between a certain user and all articles to obtain a recommendation list of the user. The invention utilizes knowledge map information to make up the sparsity of the original user historical behavior information, and describes the user and the articles from a multidimensional angle, so that the personalized recommendation result is more accurate.
Description
Technical Field
The invention relates to the field of machine learning, in particular to a knowledge graph driven personalized accurate recommendation method.
Background
With the development of information technology and the internet, people have gradually entered the era of information overload from the era of information scarcity. In this age, both information consumers and information producers face significant challenges: as information consumers, it is very difficult to find interesting information from a large amount of information. As information producers, it is also very difficult how to make the information they produce stand out and get the attention of users. Recommendation systems are an important tool to resolve this conflict. The recommendation system is used as an information filtering system, and recommends information which is most likely to be interested in a user to the user through the historical behavior, preference and other information of the user.
As the problem of information overload becomes more severe, more and more researchers are beginning to work on recommendation systems. Previously, many studies focused on optimizing collaborative filtering. Collaborative filtering techniques predict a user's likely future behavior through interactions between the user and an item. Although collaborative filtering is simple and efficient in many cases, recommendations are not effective in situations where user and item interactions are sparse. To address this problem, researchers use auxiliary information to more accurately characterize users and objects, making up for the problem of data sparseness. More and more researchers choose to use knowledge-graphs as auxiliary information. The knowledge graph has three advantages compared with other auxiliary information. The knowledge graph contains rich semantic information including potential interests of the user, which contributes to the accuracy of the recommendation result. Different relation links in the knowledge graph enable data to be denser, recommendation results are easy to diverge, and diversity of the recommendation results is facilitated. The recommendation result in the knowledge graph has a link relation with the previous historical behaviors of the user, and meanwhile, the knowledge graph can infer certain interests of the user according to the historical behaviors of the user, so that certain interpretability is brought to the recommendation result.
Knowledge-graph-driven recommendation systems based on Graph Neural Networks (GNNs) have emerged since 2019. The GNN-based method enables the model to learn the expression of the nodes through end-to-end training, makes full use of semantic and structural information in the knowledge graph, and overcomes the defect that the prior method extracts the features manually. Currently, GNN-based approaches mainly extend the architecture of graph-convolved networks (GCNs) to knowledge-graph driven recommendation systems. While these models have proven effective on some public data sets, the interaction of the nodes in the models is insufficient, making the representation of the nodes in the knowledge-graph less accurate.
Disclosure of Invention
Aiming at the problem that the node representation of the current knowledge graph-driven recommendation model based on the graph neural network is not accurate enough, the invention provides a multi-dimensional interactive knowledge graph spectrogram neural network model, which adds user features in the message transmission process and increases the interaction between neighbor nodes and a central node in the aggregation process so as to adjust the update direction of the node representation. The method can more fully utilize the information of the knowledge graph, accurately represent the characteristics of the nodes and improve the accuracy of the recommendation result.
A knowledge graph driven personalized accurate recommendation method comprises the following steps:
s1, acquiring related knowledge of the articles from a knowledge base according to the historical behaviors of the user, and constructing a knowledge map;
s2, initializing vector representation of each node and connection for the constructed knowledge graph, and determining the receptive field of the node;
s3, generating training samples according to the historical behaviors of the users, and initializing vector representations of all the users and articles;
s4, for each training sample, acquiring the receptive field of the corresponding entity of the article in the training sample in the knowledge graph, and inputting the receptive field and the sample as a graph neural network model to obtain a predicted value of the possibility of interaction between the user and the article; optimizing the model parameters by minimizing a loss function;
and S5, after the model optimization process is finished, sequencing the possibility predicted values of interaction between a certain user and all articles to obtain a recommendation list of the user.
Further, in the knowledge graph constructed in S1, the article and the attribute of the article are both used as their entity nodes.
Further, in a method for recommending articles individually and accurately driven by a knowledge graph, the knowledge related to the articles in S1 constitutes the knowledge graph in a triple < h, r, t >.
Further, in the method for recommending individual precision driven by the knowledge graph, the vector representation method for initializing all nodes and connections in S2 is Xavier initialization.
Further, a method for determining a perceptual domain of each node in S2 includes:
s21: determining one-hop neighbors of all nodes according to the knowledge graph;
s22: according to the neighbor size k determined in advance, each node randomly selects k one-hop neighbors, and if the number of the one-hop neighbors is less than k, the selected neighbors are randomly and repeatedly appeared until the number of the neighbors reaches k;
s23: forming a receptive field of each node according to the depth h of the receptive field determined in advance; for each node, the one-hop neighbor of the one-hop neighbor becomes a two-hop neighbor of the node, the one-hop neighbor of the two-hop neighbor becomes a three-hop neighbor of the node, and so on until the h-hop neighbor, all the neighbors become the receptive field of the node.
Further, a method for generating a training sample in S3 includes: for each user's historical behavior, the corresponding user-item link < u, v,1> is taken as a positive sample, while a corresponding number of negative samples < u, v', 0> are generated.
Further, in the method for recommending knowledge-graph-driven personalized precision, the method for initializing vector representation of the user and the article in S3 is Xavier initialization.
Further, in S4, the method for accurately recommending knowledge-graph-driven personalization includes the following steps:
s41, transmitting the characteristics of all nodes to adjacent nodes through connection, and simultaneously considering the connection type r and the user characteristics u in the transmission process;
s42, for each node, combining the feature sets propagated by all the one-hop neighbors into the neighborhood features, multiplying the neighborhood features by the feature element levels of the node, weighting and summing the result with the domain features and the node features, and inputting the result into an aggregation function to obtain the new features of each node;
s43, repeating S41 to S42, and taking the new characteristics of the corresponding entity of the article after repeating for h-1 times as vector representation of the article;
s44, calculating the possibility of the interaction between the item and the user through a prediction function, wherein the prediction function is the inner product of the vector representation of the item and the vector representation of the user;
and S45, calculating a loss function, minimizing the loss function through an Adam optimization algorithm, and updating the model parameters.
The invention has the beneficial effects that: according to the invention, knowledge map information is utilized to make up for the sparsity of the original user historical behavior information, and the user and the article are depicted from a multi-dimensional angle, so that the personalized recommendation result is more accurate; the invention changes the updating direction represented by the nodes by increasing the interaction between the nodes in the message transmission process and the aggregation process, so that the node updating direction meets the following four common senses: users who purchase the same item are closer, items purchased by the same user are closer, items with the same attribute are closer, and users with the same interest are closer. This makes the model some interpretability and the node representation more accurate. The invention provides a model architecture capable of end-to-end training, which does not need to manually extract features, does not introduce artificial bias into the model, and reduces manual design processes.
Drawings
FIG. 1 is a flow chart of a method for accurate recommendation of knowledge-graph driven personalization in accordance with the present invention;
FIG. 2 is a graph neural network computation process in a knowledge-graph-driven personalized precision recommendation method;
FIG. 3 is a weight calculation of an edge in a knowledge-graph-driven personalized accurate recommendation method in a message propagation process;
FIG. 4 is a new state generation process of a central node in an aggregation process of a knowledge graph-driven personalized accurate recommendation method.
Detailed Description
The technical solution of the present invention is further described below with reference to the accompanying drawings.
A knowledge graph driven personalized accurate recommendation method comprises the following steps:
s1, acquiring related knowledge of the articles from a knowledge base according to the historical behaviors of the user, and constructing a knowledge map;
s2, initializing vector representation of each node and connection for the constructed knowledge graph, and determining the receptive field of the node;
s3, generating training samples according to the historical behaviors of the users, and initializing vector representations of all the users and articles;
s4, for each training sample, acquiring the receptive field of the corresponding entity of the article in the training sample in the knowledge graph, and inputting the receptive field and the sample as a graph neural network model to obtain a predicted value of the possibility of interaction between the user and the article; optimizing the model parameters by minimizing a loss function;
and S5, after the model optimization process is finished, sequencing the possibility predicted values of interaction between a certain user and all articles to obtain a recommendation list of the user.
The technical solution of the present invention is described in detail below:
and S1, acquiring the related knowledge of the articles from the knowledge base according to the historical behaviors of the user, and constructing a knowledge graph.
In the constructed knowledge graph, the articles and the attributes of the articles are used as the entity nodes of the articles; the knowledge associated with the item constitutes a knowledge graph in the form of a triplet < h, r, t >. For example, there is a known knowledge that "chequer directed me and my home", which can be converted into the triplets < me and my home, director, chequer >.
S2, initializing vector representation of each node and connection and determining the receptive field of the node for the constructed knowledge-graph.
The vector representation method for initializing all nodes and connections is Xavier initialization.
The method for determining the receptive field of each node specifically comprises the following steps:
s21: determining one-hop neighbors of all nodes according to the knowledge graph;
s22: according to the neighbor size k determined in advance, each node randomly selects k one-hop neighbors, and if the number of the one-hop neighbors is less than k, the selected neighbors are randomly and repeatedly appeared until the number of the neighbors reaches k;
s23: forming a receptive field of each node according to the depth h of the receptive field determined in advance; for each node, the one-hop neighbor of the one-hop neighbor becomes a two-hop neighbor of the node, the one-hop neighbor of the two-hop neighbor becomes a three-hop neighbor of the node, and so on until the h-hop neighbor, all the neighbors become the receptive field of the node.
And S3, generating training samples according to the historical behaviors of the users, and initializing vector representations of all the users and the articles.
The method for generating the training sample comprises the following steps: for each user's historical behavior, the corresponding user-item link < u, v,1> is taken as a positive sample, while a corresponding number of negative samples < u, v', 0> are generated. The method for initializing the vector representation of the user and the article is Xavier initialization.
S4, for each training sample, acquiring the receptive field of the corresponding entity of the article in the training sample in the knowledge graph, and inputting the receptive field and the sample as a graph neural network model to obtain a predicted value of the possibility of interaction between the user and the article; by minimizing the loss function, the model parameters are optimized.
The graph neural network model operation comprises the following steps:
s41, transmitting the characteristics of all nodes to adjacent nodes through connection, and simultaneously considering the connection type r and the user characteristics u in the transmission process; in this process, the weight of each connection is:
in the formula, u is represented by a user, v is represented by a node corresponding to an article in a knowledge graph, vN is represented by a certain neighbor node of the node, r is represented by a vector connected with v and vN, and nv (i) is a perception domain set of the node v.
S42, for each node, combining the feature sets propagated by all the one-hop neighbors into the neighborhood features, multiplying the neighborhood features by the feature element levels of the node, weighting and summing the result with the domain features and the node features, and inputting the result into an aggregation function to obtain the new features of each node;
the neighborhood feature is represented as:
the aggregation function is represented as:
agt=σ(W1(v+puv)+W2Q(v,puv)+b)。
s43, repeating S41 to S42, and taking the new characteristics of the corresponding entity of the article after repeating for h-1 times as vector representation of the article;
s44, calculating the possibility of the interaction between the item and the user through a prediction function, wherein the prediction function is the inner product of the vector representation of the item and the vector representation of the user; the inner product is expressed as:
s45, calculating a loss function, minimizing the loss function through an Adam optimization algorithm, and updating model parameters; the loss function is:
and S5, after the model optimization process is finished, sequencing the possibility predicted values of interaction between a certain user and all articles to obtain a recommendation list of the user.
Claims (8)
1. A knowledge graph driven personalized accurate recommendation method is characterized in that: the method comprises the following steps:
s1, acquiring related knowledge of the articles from a knowledge base according to the historical behaviors of the user, and constructing a knowledge map;
s2, initializing vector representation of each node and connection for the constructed knowledge graph, and determining the receptive field of the node;
s3, generating training samples according to the historical behaviors of the users, and initializing vector representations of all the users and articles;
s4, for each training sample, acquiring the receptive field of the corresponding entity of the article in the training sample in the knowledge graph, and inputting the receptive field and the sample as a graph neural network model to obtain a predicted value of the possibility of interaction between the user and the article; optimizing the model parameters by minimizing a loss function;
and S5, after the model optimization process is finished, sequencing the possibility predicted values of interaction between a certain user and all articles to obtain a recommendation list of the user.
2. The method for knowledge-graph-driven personalized accurate recommendation according to claim 1, characterized in that: in the knowledge graph constructed in the S1, the item and the attribute of the item are both used as the entity node.
3. The method for knowledge-graph-driven personalized accurate recommendation according to claim 1, characterized in that: the knowledge associated with the item constitutes a knowledge graph in the form of a triplet < h, r, t >.
4. The method for knowledge-graph-driven personalized accurate recommendation according to claim 1, characterized in that: the vector representation method for initializing all nodes and connections in S2 is Xavier initialization.
5. The method for knowledge-graph-driven personalized accurate recommendation according to claim 1, characterized in that: the method for determining the receptive field of each node in S2 specifically includes:
s21: determining one-hop neighbors of all nodes according to the knowledge graph;
s22: according to the neighbor size k determined in advance, each node randomly selects k one-hop neighbors, and if the number of the one-hop neighbors is less than k, the selected neighbors are randomly and repeatedly appeared until the number of the neighbors reaches k;
s23: forming a receptive field of each node according to the depth h of the receptive field determined in advance; for each node, the one-hop neighbor of the one-hop neighbor becomes a two-hop neighbor of the node, the one-hop neighbor of the two-hop neighbor becomes a three-hop neighbor of the node, and so on until the h-hop neighbor, all the neighbors become the receptive field of the node.
6. The method for knowledge-graph-driven personalized accurate recommendation according to claim 1, characterized in that: the method for generating the training sample in S3 includes: for each user's historical behavior, the corresponding user-item link < u, v,1> is taken as a positive sample, while a corresponding number of negative samples < u, v', 0> are generated.
7. The method for knowledge-graph-driven personalized accurate recommendation according to claim 1, characterized in that: the method for initializing the vector representation of the user and the article in S3 is Xavier initialization.
8. The method for knowledge-graph-driven personalized accurate recommendation according to claim 1, characterized in that: in S4, the operation of the neural network model includes the following steps:
s41, transmitting the characteristics of all nodes to adjacent nodes through connection, and simultaneously considering the connection type r and the user characteristics u in the transmission process;
s42, for each node, combining the feature sets propagated by all the one-hop neighbors into the neighborhood features, multiplying the neighborhood features by the feature element levels of the node, weighting and summing the result with the domain features and the node features, and inputting the result into an aggregation function to obtain the new features of each node;
s43, repeating S41 to S42, and taking the new characteristics of the corresponding entity of the article after repeating for h-1 times as vector representation of the article;
s44, calculating the possibility of the interaction between the item and the user through a prediction function, wherein the prediction function is the inner product of the vector representation of the item and the vector representation of the user;
and S45, calculating a loss function, minimizing the loss function through an Adam optimization algorithm, and updating the model parameters.
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911183454.6A CN110955834B (en) | 2019-11-27 | 2019-11-27 | Knowledge graph driven personalized accurate recommendation method |
PCT/CN2020/117181 WO2021103789A1 (en) | 2019-11-27 | 2020-09-23 | Personalized accurate recommendation method driven by knowledge graph |
US17/726,461 US20220245488A1 (en) | 2019-11-27 | 2022-04-21 | Accurate and personalized recommendation method based on knowledge graph |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911183454.6A CN110955834B (en) | 2019-11-27 | 2019-11-27 | Knowledge graph driven personalized accurate recommendation method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110955834A true CN110955834A (en) | 2020-04-03 |
CN110955834B CN110955834B (en) | 2023-07-07 |
Family
ID=69976996
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911183454.6A Active CN110955834B (en) | 2019-11-27 | 2019-11-27 | Knowledge graph driven personalized accurate recommendation method |
Country Status (3)
Country | Link |
---|---|
US (1) | US20220245488A1 (en) |
CN (1) | CN110955834B (en) |
WO (1) | WO2021103789A1 (en) |
Cited By (22)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111507796A (en) * | 2020-04-07 | 2020-08-07 | 桂林电子科技大学 | Online shopping mall commodity recommendation method based on knowledge graph |
CN111539786A (en) * | 2020-04-15 | 2020-08-14 | 清华大学 | Conditional attention network and application method and device thereof in personalized recommendation |
CN111967972A (en) * | 2020-08-18 | 2020-11-20 | 中国银行股份有限公司 | Financial product recommendation method and device |
CN111986006A (en) * | 2020-09-02 | 2020-11-24 | 深圳壹账通智能科技有限公司 | Product recommendation method and device based on knowledge graph, computer equipment and storage medium |
CN112085559A (en) * | 2020-08-18 | 2020-12-15 | 山东大学 | Interpretable commodity recommendation method and system based on time-sequence knowledge graph |
CN112115358A (en) * | 2020-09-14 | 2020-12-22 | 中国船舶重工集团公司第七0九研究所 | Personalized recommendation method using multi-hop path features in knowledge graph |
CN112232925A (en) * | 2020-11-02 | 2021-01-15 | 哈尔滨工程大学 | Method for carrying out personalized recommendation on commodities by fusing knowledge maps |
CN112398819A (en) * | 2020-11-02 | 2021-02-23 | 杭州海康威视数字技术股份有限公司 | Method and device for recognizing abnormality |
CN112487199A (en) * | 2020-11-24 | 2021-03-12 | 杭州电子科技大学 | User characteristic prediction method based on user purchasing behavior |
CN112559864A (en) * | 2020-12-14 | 2021-03-26 | 西安电子科技大学 | Bilinear graph network recommendation method and system based on knowledge graph enhancement |
CN112632296A (en) * | 2020-12-31 | 2021-04-09 | 上海交通大学 | Knowledge graph-based paper recommendation method and system with interpretability and terminal |
CN112667885A (en) * | 2020-12-04 | 2021-04-16 | 四川长虹电器股份有限公司 | Matrix decomposition collaborative filtering method and system for coupling social trust information |
CN112801751A (en) * | 2021-02-04 | 2021-05-14 | 重庆邮电大学 | Personalized scenic spot recommendation method of multitask graph neural network |
WO2021103789A1 (en) * | 2019-11-27 | 2021-06-03 | 西北工业大学 | Personalized accurate recommendation method driven by knowledge graph |
CN112950321A (en) * | 2021-03-10 | 2021-06-11 | 北京汇钧科技有限公司 | Article recommendation method and device |
CN113407834A (en) * | 2021-06-18 | 2021-09-17 | 北京工业大学 | Knowledge graph-assisted user multi-dimensional interest extraction method |
CN113742531A (en) * | 2020-05-27 | 2021-12-03 | 杭州海康威视数字技术股份有限公司 | Graph recommendation method and device and electronic equipment |
WO2022066587A1 (en) * | 2020-09-22 | 2022-03-31 | Visa International Service Association | Graph learning-based system with updated vectors |
CN114265982A (en) * | 2021-12-22 | 2022-04-01 | 北京中科闻歌科技股份有限公司 | Intelligent recommendation method based on knowledge graph |
CN114491055A (en) * | 2021-12-10 | 2022-05-13 | 浙江辰时科技集团有限公司 | Recommendation algorithm based on knowledge graph |
CN116051247A (en) * | 2023-03-23 | 2023-05-02 | 新立讯科技股份有限公司 | Multi-mode knowledge graph-based agriculture, forestry and animal husbandry product recommendation method and system |
CN113780097B (en) * | 2021-08-17 | 2023-12-01 | 北京数慧时空信息技术有限公司 | Cultivated land extraction method based on knowledge graph and deep learning |
Families Citing this family (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115496174B (en) * | 2021-06-18 | 2023-09-26 | 中山大学 | Method for optimizing network representation learning, model training method and system |
CN113901319B (en) * | 2021-10-18 | 2024-03-26 | 桂林电子科技大学 | Site recommendation method based on sequence semantics and attribute map feature learning |
CN114169975A (en) * | 2021-12-17 | 2022-03-11 | 福州大学 | Shopping network commodity recommendation method and system based on random walk heterogeneous attention |
CN114969508A (en) * | 2022-04-27 | 2022-08-30 | 阿里巴巴达摩院(杭州)科技有限公司 | Recommendation model data processing method |
CN115618108B (en) * | 2022-10-24 | 2023-05-26 | 云南中烟工业有限责任公司 | Brand recommendation method based on knowledge graph in new retail model |
CN115564314A (en) * | 2022-11-24 | 2023-01-03 | 安徽蓝海之光科技有限公司 | Regional carbon emission intelligent measurement system based on low-carbon energy consumption optimization cooperation |
CN116402589B (en) * | 2023-06-05 | 2023-09-15 | 湖南师范大学 | Commodity recommendation method and system based on knowledge graph and rotary coding |
CN117094762B (en) * | 2023-10-17 | 2024-01-23 | 北京中奥通宇科技股份有限公司 | User rights and interests recommending method and system based on AI artificial intelligence |
CN117312542B (en) * | 2023-11-29 | 2024-02-13 | 泰山学院 | Reading recommendation method and system based on artificial intelligence |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108733798A (en) * | 2018-05-17 | 2018-11-02 | 电子科技大学 | A kind of personalized recommendation method of knowledge based collection of illustrative plates |
CN108804633A (en) * | 2018-06-01 | 2018-11-13 | 腾讯科技(深圳)有限公司 | The content recommendation method of Behavior-based control Semantic knowledge network |
CN108920527A (en) * | 2018-06-07 | 2018-11-30 | 桂林电子科技大学 | A kind of personalized recommendation method of knowledge based map |
US20190034512A1 (en) * | 2017-07-26 | 2019-01-31 | General Electric Company | Subject matter expert knowledge mapping using dynamic clustering |
CN110321482A (en) * | 2019-06-11 | 2019-10-11 | 阿里巴巴集团控股有限公司 | A kind of recommended method of information, device and equipment |
CN110489547A (en) * | 2019-07-11 | 2019-11-22 | 桂林电子科技大学 | A kind of tourist attractions recommended method and device based on hybrid supervised learning |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110955834B (en) * | 2019-11-27 | 2023-07-07 | 西北工业大学 | Knowledge graph driven personalized accurate recommendation method |
-
2019
- 2019-11-27 CN CN201911183454.6A patent/CN110955834B/en active Active
-
2020
- 2020-09-23 WO PCT/CN2020/117181 patent/WO2021103789A1/en active Application Filing
-
2022
- 2022-04-21 US US17/726,461 patent/US20220245488A1/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20190034512A1 (en) * | 2017-07-26 | 2019-01-31 | General Electric Company | Subject matter expert knowledge mapping using dynamic clustering |
CN108733798A (en) * | 2018-05-17 | 2018-11-02 | 电子科技大学 | A kind of personalized recommendation method of knowledge based collection of illustrative plates |
CN108804633A (en) * | 2018-06-01 | 2018-11-13 | 腾讯科技(深圳)有限公司 | The content recommendation method of Behavior-based control Semantic knowledge network |
CN108920527A (en) * | 2018-06-07 | 2018-11-30 | 桂林电子科技大学 | A kind of personalized recommendation method of knowledge based map |
CN110321482A (en) * | 2019-06-11 | 2019-10-11 | 阿里巴巴集团控股有限公司 | A kind of recommended method of information, device and equipment |
CN110489547A (en) * | 2019-07-11 | 2019-11-22 | 桂林电子科技大学 | A kind of tourist attractions recommended method and device based on hybrid supervised learning |
Non-Patent Citations (4)
Title |
---|
WEI CHEN等: "Jointing Knowledge Graph and Neural Network for TOP-N Recommendation", 《SPRINGER LINK》 * |
XIANG WANG等: "KGAT:Knowledge Graph Attention Networks for Recommendation", 《ARXIV》 * |
孙文珺: "基于深度学习模型的感应电机故障诊断方法研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
靳京: "基于深度学习融入实体描述的知识图谱表示学习研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (30)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2021103789A1 (en) * | 2019-11-27 | 2021-06-03 | 西北工业大学 | Personalized accurate recommendation method driven by knowledge graph |
CN111507796A (en) * | 2020-04-07 | 2020-08-07 | 桂林电子科技大学 | Online shopping mall commodity recommendation method based on knowledge graph |
CN111539786A (en) * | 2020-04-15 | 2020-08-14 | 清华大学 | Conditional attention network and application method and device thereof in personalized recommendation |
CN111539786B (en) * | 2020-04-15 | 2022-05-06 | 清华大学 | Conditional attention network and application method and device thereof in personalized recommendation |
CN113742531B (en) * | 2020-05-27 | 2023-09-01 | 杭州海康威视数字技术股份有限公司 | Picture recommendation method and device and electronic equipment |
CN113742531A (en) * | 2020-05-27 | 2021-12-03 | 杭州海康威视数字技术股份有限公司 | Graph recommendation method and device and electronic equipment |
CN112085559A (en) * | 2020-08-18 | 2020-12-15 | 山东大学 | Interpretable commodity recommendation method and system based on time-sequence knowledge graph |
CN111967972B (en) * | 2020-08-18 | 2024-02-27 | 中国银行股份有限公司 | Financial product recommendation method and device |
CN111967972A (en) * | 2020-08-18 | 2020-11-20 | 中国银行股份有限公司 | Financial product recommendation method and device |
CN111986006A (en) * | 2020-09-02 | 2020-11-24 | 深圳壹账通智能科技有限公司 | Product recommendation method and device based on knowledge graph, computer equipment and storage medium |
CN112115358B (en) * | 2020-09-14 | 2024-04-16 | 中国船舶重工集团公司第七0九研究所 | Personalized recommendation method utilizing multi-hop path characteristics in knowledge graph |
CN112115358A (en) * | 2020-09-14 | 2020-12-22 | 中国船舶重工集团公司第七0九研究所 | Personalized recommendation method using multi-hop path features in knowledge graph |
WO2022066587A1 (en) * | 2020-09-22 | 2022-03-31 | Visa International Service Association | Graph learning-based system with updated vectors |
CN112398819A (en) * | 2020-11-02 | 2021-02-23 | 杭州海康威视数字技术股份有限公司 | Method and device for recognizing abnormality |
CN112232925A (en) * | 2020-11-02 | 2021-01-15 | 哈尔滨工程大学 | Method for carrying out personalized recommendation on commodities by fusing knowledge maps |
CN112487199A (en) * | 2020-11-24 | 2021-03-12 | 杭州电子科技大学 | User characteristic prediction method based on user purchasing behavior |
CN112667885B (en) * | 2020-12-04 | 2022-08-16 | 四川长虹电器股份有限公司 | Matrix decomposition collaborative filtering method and system for coupling social trust information |
CN112667885A (en) * | 2020-12-04 | 2021-04-16 | 四川长虹电器股份有限公司 | Matrix decomposition collaborative filtering method and system for coupling social trust information |
CN112559864B (en) * | 2020-12-14 | 2023-03-31 | 西安电子科技大学 | Bilinear graph network recommendation method and system based on knowledge graph enhancement |
CN112559864A (en) * | 2020-12-14 | 2021-03-26 | 西安电子科技大学 | Bilinear graph network recommendation method and system based on knowledge graph enhancement |
CN112632296B (en) * | 2020-12-31 | 2022-09-23 | 上海交通大学 | Knowledge graph-based paper recommendation method and system with interpretability and terminal |
CN112632296A (en) * | 2020-12-31 | 2021-04-09 | 上海交通大学 | Knowledge graph-based paper recommendation method and system with interpretability and terminal |
CN112801751A (en) * | 2021-02-04 | 2021-05-14 | 重庆邮电大学 | Personalized scenic spot recommendation method of multitask graph neural network |
CN112950321A (en) * | 2021-03-10 | 2021-06-11 | 北京汇钧科技有限公司 | Article recommendation method and device |
CN113407834A (en) * | 2021-06-18 | 2021-09-17 | 北京工业大学 | Knowledge graph-assisted user multi-dimensional interest extraction method |
CN113780097B (en) * | 2021-08-17 | 2023-12-01 | 北京数慧时空信息技术有限公司 | Cultivated land extraction method based on knowledge graph and deep learning |
CN114491055A (en) * | 2021-12-10 | 2022-05-13 | 浙江辰时科技集团有限公司 | Recommendation algorithm based on knowledge graph |
CN114265982B (en) * | 2021-12-22 | 2022-08-26 | 北京中科闻歌科技股份有限公司 | Intelligent recommendation method based on knowledge graph |
CN114265982A (en) * | 2021-12-22 | 2022-04-01 | 北京中科闻歌科技股份有限公司 | Intelligent recommendation method based on knowledge graph |
CN116051247A (en) * | 2023-03-23 | 2023-05-02 | 新立讯科技股份有限公司 | Multi-mode knowledge graph-based agriculture, forestry and animal husbandry product recommendation method and system |
Also Published As
Publication number | Publication date |
---|---|
US20220245488A1 (en) | 2022-08-04 |
WO2021103789A1 (en) | 2021-06-03 |
CN110955834B (en) | 2023-07-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110955834B (en) | Knowledge graph driven personalized accurate recommendation method | |
CN111428147B (en) | Social recommendation method of heterogeneous graph volume network combining social and interest information | |
CN111222054B (en) | Session social contact recommendation method based on context neighbor relation modeling | |
CN112232925A (en) | Method for carrying out personalized recommendation on commodities by fusing knowledge maps | |
Chen et al. | A group recommendation system with consideration of interactions among group members | |
CN112613602A (en) | Recommendation method and system based on knowledge-aware hypergraph neural network | |
CN108563755A (en) | A kind of personalized recommendation system and method based on bidirectional circulating neural network | |
CN109190030B (en) | Implicit feedback recommendation method fusing node2vec and deep neural network | |
Aghdam et al. | A novel non-negative matrix factorization method for recommender systems | |
CN108629671B (en) | Restaurant recommendation method integrating user behavior information | |
CN112801749B (en) | Interest point recommendation method fusing context characteristics of continuous sign-in behavior sequence of user | |
CN114510653B (en) | Social group recommendation method, system, device and storage medium | |
Shafqat et al. | A hybrid GAN-based approach to solve imbalanced data problem in recommendation systems | |
CN115114421A (en) | Question-answer model training method | |
CN110008411B (en) | Deep learning interest point recommendation method based on user sign-in sparse matrix | |
CN113487018A (en) | Global context enhancement graph neural network method based on session recommendation | |
CN106407379A (en) | Hadoop platform based movie recommendation method | |
CN117495511A (en) | Commodity recommendation system and method based on contrast learning and community perception | |
CN105868422A (en) | Collaborative filtering recommendation method based on elastic dimensional feature vector optimized extraction | |
Baker et al. | Machine learning: factorization machines and normalized discounted cumulative gain for tourism recommender system optimisation | |
CN116821519A (en) | Intelligent recommendation method for system filtering and noise reduction based on graph structure | |
CN116187852A (en) | Online course recommendation method based on community association and behavior feature learning | |
CN108830680A (en) | Personalized recommendation method, system and storage medium based on discrete disassembler | |
CN113313381B (en) | User interaction sensitive dynamic graph sequence recommendation system | |
CN115408505A (en) | Dialogue recommendation algorithm based on two-channel hypergraph interest modeling |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |